XB-ART-49512PLoS One January 1, 2014; 9 (10): e110559.
FoxA4 favours notochord formation by inhibiting contiguous mesodermal fates and restricts anterior neural development in Xenopus embryos.
In vertebrates, the embryonic dorsal midline is a crucial signalling centre that patterns the surrounding tissues during development. Members of the FoxA subfamily of transcription factors are expressed in the structures that compose this centre. Foxa2 is essential for dorsal midline development in mammals, since knock-out mouse embryos lack a definitive node, notochord and floor plate. The related gene foxA4 is only present in amphibians. Expression begins in the blastula -chordin and -noggin expressing centre (BCNE) and is later restricted to the dorsal midline derivatives of the Spemann''s organiser. It was suggested that the early functions of mammalian foxa2 are carried out by foxA4 in frogs, but functional experiments were needed to test this hypothesis. Here, we show that some important dorsal midline functions of mammalian foxa2 are exerted by foxA4 in Xenopus. We provide new evidence that the latter prevents the respecification of dorsal midline precursors towards contiguous fates, inhibiting prechordal and paraxial mesoderm development in favour of the notochord. In addition, we show that foxA4 is required for the correct regionalisation and maintenance of the central nervous system. FoxA4 participates in constraining the prospective rostral forebrain territory during neural specification and is necessary for the correct segregation of the most anterior ectodermal derivatives, such as the cement gland and the pituitary anlagen. Moreover, the early expression of foxA4 in the BCNE (which contains precursors of the whole forebrain and most of the midbrain and hindbrain) is directly required to restrict anterior neural development.
PubMed ID: 25343614
PMC ID: PMC4208771
Article link: PLoS One
Genes referenced: ag1 chrd.1 egr1 egr2 emx1 emx1l en2 foxa2 foxa4 frzb gsc hes4 hesx1 hoxb1 hoxb7 hoxc6 myf5 myod1 nog otx2 shh six3 slc5a5 sox2 tbxt tubb2b
Morpholinos: foxa4 MO1
Article Images: [+] show captions
|FIGURE 1: FoxA4 depletion altered the expression of mesodermal markers, impaired axis elongation, expanded the prechordal mesoderm and reduced the notochord. (A–J) Expression of mesodermal markers at stage 13, dorsal views. (A, F) myod. (B, G) bra. (C, H) gsc. (D, I) frzb1. (E, J) chd. Embryos were injected into one dorsal cell at the 4-cell stage with 40 ng of CtrlMO (A) or of FoxA4MO (B), or with 20 ng of each MO into both dorsal cells at the 4-cell stage, (CtrlMO: B,C,E; FoxA4MO: G, H, J) or at the 1-cell stage (CtrlMO: D; FoxA4MO: I). The myoD domain was shortened on the injected side (is) by FoxA4MO (58%, n = 12) (F). The bra domain was split by FoxA4MO (G) (79%, n = 24). The gsc domain was posteriorly expanded by FoxA4MO (100%, n = 15) (H). The frzb1 domain was posteriorly expanded by FoxA4MO (100%, n = 14) (I). The chd domain was affected in 100% of the injected embryos (n = 14) (J). Among these, 71% showed an expansion of the anterior chd domain, corresponding to the PM (green arrow), while the posterior domain (corresponding to the notochord) was severely reduced (red arrow). In the remaining 29%, the entire axial mesoderm appeared much thinner and shorter than in CtrlMO-injected siblings (inset in J). Similar results were obtained after injection of 20 ng of FoxA4MO into 1-cell stage embryos (Fig. S1B). (K) The body length between the cement gland and the tip of the tailbud was measured at tailbud stage in embryos injected before the first cleavage with 20 ng of CtrlMO (green bar, n = 25) or of FoxA4MO (red bar, n = 27). P<0.0001, two-tailed t-test. (L) Notochord/embryo length ratio measured in the group of embryos corresponding to those shown in (B, G). Green bar, CtrlMO (n = 18); red bar, FoxA4MO (n = 24). P<0.0001, two-tailed t-test. (M–R) Expression of axial mesodermal markers at stage 10.25. (M, P) gsc. (N, Q) frzb1. (O, R) chd. Embryos were injected with 20 ng of CtrlMO (M–O) or of FoxA4MO (P–R). Injections were performed unilaterally in 1 cell at the 2 or 4-cell stage or at the 1-cell stage, giving similar results. Unilateral injections are shown here. The injected side was identified by DOG fluorescence and is oriented to the right (insets in M–R). In gastrulae, we were able to observe an expansion of gsc (58%, n = 40; P) and of frzb1 domains (88%, n = 33; Q). chd expression was reduced in the less involuted, superficial cells on the injected side (red arrows), but a cloud of deep cells expressing chd persisted (green arrows), and even appeared to be expanded on the injected side in some embryos, like the one shown below (50%, n = 10) (R). doi:10.1371/journal.pone.0110559.g001|
|Figure 2. FoxA4MO disrupted DML formation. (A–H) Stage 15 embryos left uninjected (A) or injected with 0.5 ng of foxA4FL mRNA (E), 40 ng of CtrlMO (B–D) or of FoxA4MO (F–H) before the first cleavage. (I–T) Tailbud stage embryos injected into two dorsal cells at the 4-cell stage with 20 ng of CtrlMO or of FoxA4MO. The descendants of these cells give rise to the DML and the cephalic region. Expression of chd (A, B, E, F, I, J, M, N, Q, R), shh (C, G, K, L, O, P, S, T), hairy2 (D, H). The inset in K shows a magnified view of the area depicted by the rectangle; fp, floor plate; no, notochord; h, hypochord. Embyos injected with foxA4FL mRNA showed a thicker notochord than control siblings, as revealed by chd expression (87%, n = 15) (E). The percentage of embryos injected with FoxA4MO showing the indicated changes in the corresponding markers is indicated between parentheses, as follows. At neural plate stage, the expression of chd (90%, n = 20) (F) and shh (54%, n = 22) (G) was patchy, and hairy2 expression was reduced in the FP domain (60%, n = 23) (H). At stage 26, the chd domain was patchy and/or thinner (80%, n = 21) (M, N, Q, R). Arrowheads point to chd + cells in dorsal and ventral positions with respect to a notochordal gap. Shh was also patchy (arrowheads, O, P) or completely abolished (S,T) (86%, n = 22). In overall, tailbuds injected with FoxA4MO showed reduced heads (89%, n = 51) (M, Q, O, S). (A–H, J, L, N, P, R, T) dorsal views; (I, K, M, O, Q, S) lateral views. doi:10.1371/journal.pone.0110559.g002|
|Figure 3. The DML is invaded by paraxial mesoderm and neuroectoderm in FoxA4 morphants. Dorsal views of embryos at neural plate stage hybridised with myf5 (A, D), myoD (B, E, G, H), or sox2 probes (C, F). They were injected before the first cleavage with 20 ng CtrlMO (A–C) or of FoxA4MO (D–F), 0.25 ng of foxA4FL mRNA (H), or left uninjected (G). In FoxA4 morphants, myf5 (D) and myoD expression (E) was found in the DML (40%, n = 23; 56%, n = 23, respectively). The left embryo in E shows that myoD expression invaded the DML along the A–P axis. The right embryo in E shows stretches of the myoD domains fused at the midline (arrowheads). Sox2 transcripts obliterated the prospective FP (79%, n = 19) (F). In embryos injected with foxA4FL mRNA, myoD expression was reduced, and the medial borders of the bilateral domains were more separated (H, 42%, n = 26) than in control siblings (G). doi:10.1371/journal.pone.0110559.g003|
|Figure 4. Knock-down and overexpression of foxA4 shifted the anterior boundary of sox2 in opposite directions. Sox2 expression in stage 14–15 embryos injected with 20 ng of CtrlMO (A, E) or FoxA4MO (B, F) or with 0.5 ng of foxA4FL mRNA (D, H) into one cell at the 2 cell stage, or left uninjected (C, G). FoxA4MO produced a caudal shift (62,5%, n = 16) (B, F). FoxA4FL mRNA produced an anterior shift of the anterior boundary of sox2 (52%, n = 23) (D, H). (A–D) dorsal views; (E–H) anterior views; is, injected side; nis, non-injected side. doi:10.1371/journal.pone.0110559.g004|
|Figure 5. Xanf1 and otx2 progressively establish complementary domains. Normal expression of Xanf1, en2 and hoxb7 (A, B, C) and otx2 and krox20 (D–F) in progressively older neural plate stage embryos in anterior views. Yellow asterisks, posterior border of Xanf1 in the ANR; red asterisks, anterior border of Xanf1 in the ANR; white asterisks, SHA; black asterisks, expression hole in the neural otx2 subdomain; black arrows, caudal diencephalic-mesencephalic stripes demarcating the posterior border of the otx2 neural subdomain; yellow arrows, stripe corresponding to the anterior border of otx2 in the ANR; red arrows, SHA; white arrows, CGA. (G) Diagram summarising the expression patterns of otx2, Xanf1, en2, and krox20 in an anterior view of embryos at the stages shown in B, C, E, F. ANR, anterior neural ridge; r3, third rhombomere; m/h, midbrain/hindbrain boundary. doi:10.1371/journal.pone.0110559.g005|
|Figure 6. FoxA4 overexpression produced complementary effects on Xanf1 and otx2 and disfavoured anterior development. (A, B, F, G, K, L) Anterior views of early neural plate stage embryos analysed by WMISH with the following markers: Xanf1 and en2 (A, F, K), otx2 and krox20 (B, G, L). Embryos were injected with 0.25 ng (F, K) or 1 ng (G, L) of foxA4FL mRNA before the first cleavage, or were left uninjected (A, B). FoxA4 overexpression affected Xanf1 expression in 83% of the injected embryos (n = 23) as follows: it was reduced in the ANR but not in the SHA (48%, n = 23) (F), or it was reduced in the whole domain (35%, n = 23) (K). The expression hole of the otx2 domain was filled with otx2 transcripts (42%, n = 19) (G); the otx2 domain was reduced, but the caudal diencephalic/mesencephalic stripes were expanded (89%, n = 19) (L). (C–E, H–J, M–O) Anterior views of tailbuds (stage 27/28) injected with 0.25 ng (H, I, M, N) or 0.5 ng (J, O) of foxA4FL mRNA before the first cleavage (I, J, M–O) or into one cell at the 2-cell stage (H), or left uninjected (C, D, E). Expression of otx2 (C,H,M), Xanf1 and en2 (D, I, N), Xag1/N-tubulin (E, J, O). After overexpression of foxA4, the caudal diencephalic otx2 subdomain was expanded anteriorly (54%, n = 13) (H) or the telencephalic otx2 subdomain disappeared (40% n = 10) (M). The pituitary Xanf1 domain was expanded (33%, n = 15) (I). N-tubulin expression in the rostral forebrain (rectangle in E) was deleted and the Xag1 + cement and hatching glands were reduced (J) or the cement gland was absent (O) (54%, n = 24). The most severe phenotypes presented head truncations (23%, n = 13) (M,N,O). Red arrowhead, anterior limit of the caudal diencephalic otx2 subdomain; white arrowhead, posterior limit of the ventral telencephalic otx2 subdomain; vtel, ventral telencephalic otx2; cdi, caudal diencephalic otx2; mes, mesencephalic otx2; op, olfactory placode; epiph, epiphisis; ev, eye vesicle; cg, cement gland; hg, hatching gland cells; pit, pituitary anlage; is, injected side; nis, non-injected side. doi:10.1371/journal.pone.0110559.g006|
|Figure 7. FoxA4 inhibition affected anterior ectodermal/neural specification. Embryos at early neural plate stage injected with 20 ng of CtrlMO or of FoxA4MO before the first cleavage (A, B, F, G, K–N, P–S) or unilaterally injected into 1 cell at the 2-cell stage (C, D, H, I, O, T). The injected side was determined by DOG fluorescence (insets). Embryos were hybridised with the following probes: Xanf1, en2, and hoxb7 (A–C, F–H); hoxc6 (D, I); otx2 and krox20 (K, L, P, Q); Xag1 and N-tubulin (M, N, R, S); en2 and hoxb1 (O, T). Green arrows in (A, F) point to the anterior limit of hoxb7. At neural plate stage, Xag1 marks the cement gland anlage (indicated between red arrowheads in N), and the hatching gland primordia (red arrows in N)  . Black arrowhead in (H), caudal shift of Xanf1 and fusion with the en2 stripe. Black arrow in (H), fusion of the SHA and the ANR. Red arrow in (I), down-regulation of hoxc6 on the injected side. Yellow arrowhead in (Q), diffuse CGA and anterior border of the ANR. Notice the down-regulation and the caudal shift of en2 (red arrows) and hoxb1 (green arrows) by comparing the injected side (right) with the non-injected side (left) in the FoxA4 morphant in (T) and with the CtrlMO-injected embryo in (O). (E) Ratio between the distance from the posterior limit of Xanf1 to the blastopore and the total length of the embryo in the groups shown in A, B, F, G (r-Xanf1). The ratio was significantly lower in FoxA4 morphants; P<0.0001, two-tailed t-test. (J) The ratio between the length of the hoxb7 domain and the embryo’s length was significantly lower in FoxA4 morphants, as measured in the groups shown in A, B, F, G; P<0.0001, two-tailed t-test. ANR, anterior neural ridge; CGA, cement gland anlage; SHA, stomodeal-adenohypophyseal anlage; m/h, midbrain/hindbrain boundary; m, d, presumptive mesencephalic and caudal diencephalic regions expressing otx2; r3, third rhombomere; r4, fourth rhombomere. (A, D, F, I, K, M, O, P, Q, R, T) are dorsal views; (B, C, G, H, L, N, Q, S) are anterior views. doi:10.1371/journal.pone.0110559.g007|
|Figure 8. Depletion of FoxA4 led to head defects at tailbud stages. Anterior views of stage 27/28 embryos injected with 20 ng of CtrlMO (A–E) or FoxA4MO (G–K). Embryos were injected into one dorsal cell at 4-cell stage (A, G) or prior to the first cleavage (B–E, H–K). Expression of otx2 (A, B, G, H), six3 (C, I), emx1 (D, J) and Xanf1 (E, K). Results are summarised in F, L. In embryos unilaterally injected with FoxA4MO, the telencephalic and caudal diencephalic otx2 subdomains were fused in the injected side (red arrowhead), and the eye was lost (70%, n = 24) (G). When FoxA4MO was homogenously distributed, the otx2 domain was reduced, the telencephalic subdomain was fused in the midline and also, with the caudal diencephalic subdomain (80%, n = 20) (H). FoxA4MO reduced the expression domains of six3 (87%, n = 23) (I), emx1 (80%, n = 26) (J) and Xanf1 (70%, n = 20) (K). dtel, dorsal telencephalon; vtel, ventral telencephalic otx2; cdi, caudal diencephalic otx2; mes, mesencephalic otx2; op, olfactory placode; epiph, epiphisis; ev, eye vesicle; cg, cement gland; pit, pituitary anlage; is, injected side; nis, non-injected side. The black line in F, L represents the contour of the head. For comparison, the head contour of the CtrlMO-injected embryo was projected on the FoxA4 morphant diagram (dotted black line in L). doi:10.1371/journal.pone.0110559.g008|
|Figure 9. FoxA4 depletion altered neurogenesis along the A–P axis. (A–J) N-tubulin and Xag1 expression in stage 27/28 embryos injected with 40 ng of CtrlMO (A, C, E–G) or FoxA4MO (B, D, H–J) into one cell at the 2-cell stage. (A, B) Dorsal views. (C, D) Magnified images of the trunk region of another pair of CtrlMO (C) and FoxA4MO (D) injected embryos, shown in dorso-lateral views to facilitate comparisons. The red arrow points to an N-tubulin + projection on the injected side. (E, F, H, I) Lateral views of the head region. (G, J) Anterior views. N-tubulin is expressed in the trigeminal (tn) and vestibulocochlear (vn) nerves. In FoxA4MO-injected embryos, the tn and vn were disturbed (red arrowhead) (H), as revealed by N-tubulin; the cement gland (cg) was bent and closer to the forebrain, and the hatching gland (hg) was shortened on the FoxA4-depleted side, as revealed by Xag1 (green arrowhead) (J) (90%, n = 21). (K–N) Transverse sections at the levels indicated by dotted black lines in C,D, shown in bright field (K, M) and their corresponding nuclear Hoescht fluorescence (L, N). Bright field images were processed with Adobe Photoshop CS2 in order to superimpose the N-tubulin expression (yellow) to the Hoescht fluorescence. Yellow arrows point to the projection emerging from the dorsal neural tube; cyan arrows point to both sides of the ventral neural tube; green arrows point to the place left by the disorganised or absent notochord; no, notochord; is, injected side; nis, non-injected side. (O, P) Summary of the results shown in A–J. doi:10.1371/journal.pone.0110559.g009|
|Figure 10. Summary of the phenotypes after manipulating foxA4 function in Xenopus embryos. The diagrams show the phenotypes in embryos with normal (A, D), depleted (B, E), or excess levels (C, F) of foxA4 function, as revealed by the expression patterns of bra, chd, myf5/myoD, gsc, frzb1, otx2, Xanf1, krox20 and en2 at neural plate stage. (A–C) In DML development, foxA4 is necessary to restrict the paraxial mesodermal (PAM) and the prechordal mesodermal (PM) fates, allowing notochord (NO) development. (D–F) Controlled levels of foxA4 function are necessary for the correct specification and segregation of the anterior neural and ectodermal anlagen. ANR, anterior neural ridge; SHA, stomodeal-adenohypophyseal anlage; CGA, cement gland anlage; r3, third rhombomere; m/h, midbrain/hindbrain boundary. The asterisk in D marks the expression hole left by otx2 in the ANR, which is occupied by Xanf1. The dotted blue line (D–F) represents the contour of the neural plate. The dashed black line (D–F) represents the anterior boundary of the neural ectoderm. (A–C) Dorsal views. (D–F) Anterior views. See text for details. doi:10.1371/journal.pone.0110559.g010|
|Figure 11. FoxA4 depletion induced robust cell death in the cephalic region.TUNEL analysis of stage 28 embryos injected with 20 ng of CtrlMO or FoxA4MO prior to the first cleavage. (A–D) Anterior views. (E–H) Lateral views of the same embryos shown in A–D, cleared in Murray's solution. CtrlMO-injected embryos showed scattered apoptotic cells (blue dots) (A, E). FoxA4MO-injected siblings presented accumulation of apoptotic cells in the head region (42%, n = 12) (B–D, F–H).|
|Figure 12. FoxA4MO reproduced the effects on anterior neural markers in embryos in which mesoderm induction was blocked with Cer-S.(A–D) Expression of foxA4 in sibling controls (A, C) or in embryos injected with Cer-S mRNA (B, D), analysed at stage 9 (A, B) or at stage 11 (C, D). After blocking mesoderm induction, foxa4 expression persisted in the BCNE (B) (100%, n = 15) but was suppressed from the axial mesoderm (D) (100%, n = 7). (E, F) MyoD expression at neural plate stage in a sibling control (E) and in an embryo injected with Cer-S. Mesoderm was suppressed in 100% of the injected embryos analyzed with myoD (100%, n = 8). (G–I) Expression of Xanf1 and en2 at neural plate stage in a sibling control (G), and in embryos in which mesoderm induction was blocked with Cer-S and were injected at the 1-cell stage with either 20 ng of CtrlMO (H) or 20 ng of FoxA4MO (I). Embryos injected with CtrlMO + Cer-S showed expression of Xanf1 (yellow arrow) and en2 (black arrow) (H, 100%, n = 8). In embryos injected with FoxA4MO + Cer-S, Xanf1 was expanded and up-regulated (yellow arrow) and en2 was down-regulated (black arrow) (I, 100%, n = 10) in comparison to embryos injected with CtrlMO + Cer-S (H). (J–L) Expression of otx2 at neural plate stage in a sibling control (J), and in embryos in which mesoderm induction was blocked with Cer-S and were injected at the 1-cell stage with either 20 ng of Ctrl MO (K) or 20 ng of FoxA4MO (L). Embryos injected with CtrlMO + Cer-S showed expression of otx2 (K, 100%, n = 4). In embryos injected with FoxA4MO + Cer-S, otx2 was down-regulated (L, 100%, n = 20) in comparison to embryos injected with CtrlMO + Cer-S (K).|
|Figure 13. Depletion of FoxA4 from the ectoderm expands Xanf1 and suppresses en2 in the anterior neural plate.(A) Embryos were injected at the 1-cell stage with 20 ng of FoxA4MO +20 ng of DOG (green) or were left uninjected (grey). At stage 9, ectodermal explants were excised and the following recombinants were obtained: in Control recombinants, the ectodermal explant of a recipient uninjected embryo was replaced with the ectodermal explant from an uninjected donor. In FoxA4MO ect/Control mes recombinant, the ectoderm was provided by a FoxA4MO-injected donor (green), while the mesoderm derived from an uninjected sibling recipient (grey). In Control ect/FoxA4MO mes recombinant, the ectoderm was provided by an uninjected sibling donor (grey), while the mesoderm derived from a FoxA4MO-injected recipient (green). (B–E′) Analysis of Xanf1 and en2 when uninjected sibling controls (B) reached neurula stage. (C,C′) Control recombinant. (D–D′) FoxA4MO ect/Control mes recombinant. (E–E′) Control ect/FoxA4MO mes recombinant. (D′ and E′) DOG fluorescence of the images shown in D′, E′, respectively.|
|Figure 14. Model for foxA4 function in Xenopus embryos.(A) Summary of foxA4 functions in A–P regionalisation and in DML development. For simplicity, they are depicted in a diagram of an embryo at neural plate stage, shown in dorsal view, when foxA4 is only expressed in the DML (pink). FoxA4 modulates A–P development by inhibiting anterior fates (red lines) in the axial mesoderm (prechordal mesoderm, PM) and in the neuroectoderm (NE) and ectoderm, while favouring posterior fates (green arrows) in the neural plate and the dorsal midline (notochord, NO; floor plate, FP). In the trunk, foxA4 prevents the respecification of dorsal midline precursors towards contiguous fates, by inhibiting (red lines) the paraxial mesodermal fate (PAM). Yellow, prechordal mesoderm. Grey, paraxial mesoderm. The dotted light blue line demarcates de neural plate. (B) Diagram of a blastula stage embryo in dorsal view, showing the expression of foxA4 in the BCNE centre (pink), which is composed by precursors of the Spemann's organiser and of the whole forebrain and most of the midbrain and hindbrain . An, animal; Veg, vegetal. FoxA4 modulates the initial CNS regionalisation by operating on the BCNE, favouring posterior fates among BCNE derivatives (green arrow), while restricting anterior fates (red line), as revealed by the markers Xanf1 (rostral forebrain), otx2 (caudal forebrain/midbrain), en2 (midbrain/hindbrain boundary).|